Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2010-07-25

00:00:48 Quit ender` (Quit: Today's BOFH excuse: Zombie processes haunting the computer)
00:05:02***Saving seen data "./dancer.seen"
00:11:25 Quit RandomCake (Ping timeout: 258 seconds)
00:22:57 Quit sudoman (Quit: Leaving)
00:31:32 Join RandomCake [0] (~chatzilla@
00:41:48CIA-8New commit by 03bieber (r27543): Theme Editor: Added resource check to project export
00:43:31CIA-8r27543 build result: All green
01:01:33 Join BlakeJohnson861 [0] (
01:02:42 Join BlakeJohnson862 [0] (
01:05:41 Quit BlakeJohnson86 (Ping timeout: 276 seconds)
01:05:49 Quit BlakeJohnson861 (Ping timeout: 245 seconds)
01:05:52 Quit Jerom (Read error: Connection reset by peer)
01:08:30 Join bunnyboi [0] (
01:22:05pixelmathe gigabeat S can only record from the radio?
01:37:14 Join Drise [0] (
01:37:57DriseHey, I was told to come to here to be registered as WikiUserGroup
01:38:20DriseWikiname : RichMoll
01:39:26DriseIf anyone could hep that would be great, please and thank you in advanced
01:40:12saratogaDrise: done
01:40:21DriseThank you again.
01:52:37 Join webguest31 [0] (
01:53:53 Quit webguest31 (Client Quit)
01:55:50DriseWhile I'm here, anything new done on the Fuze V2 USB? Any updates since like a week and a half ago?
01:57:28 Quit rasher (Ping timeout: 248 seconds)
01:58:52 Join Rockstruck [0] (
01:59:06 Quit bmbl (Quit: Bye!)
02:00:14RockstruckHello, I am having difficulty setting up the rockbox development environment as described on:
02:00:32RockstruckThe problem is that after I use svn to get the files
02:00:46Rockstruckand I run configure
02:00:50Rockstruckit gives me this error:
02:01:33Rockstruck../tools/configure: line 1: arm-elf-eabi-gcc: command not found
02:01:58RockstruckI ran into the same problem when I tried the cygwin way of doing it
02:02:25Rockstruckas described on:
02:02:58RockstruckDo you know why that is happening, and how I can remedy it?
02:04:50DriseWell, I'm not sure of the problem, but it refers to line 1... What is line 1 doing or what does it contain? Is there a way to access this?
02:05:04Drise** Can you access it?
02:05:07***Saving seen data "./dancer.seen"
02:05:11 Join rasher [0] (
02:05:11 Quit rasher (Changing host)
02:05:11 Join rasher [0] (~rasher@rockbox/developer/rasher)
02:05:35RockstruckI believe that's line 1 of a script
02:05:38Rockstruckbut I'm not sure
02:06:22RockstruckI can try to redirect it to a log and upload it to pastebin or something if you'd like
02:06:29DriseCan you get into it and see what line 1 is?
02:06:44RockstruckI'll see, hold on.
02:07:35 Join JdGordon1 [0] (~jonno@
02:11:30RockstruckThe only place I see arm-elf-eabi- mentioned is on line 45 of rockbox/tools/configure
02:11:49Rockstruckwhere it appears to be deciding which tools to use
02:12:07Rockstruckand calling "prefixtools"
02:13:28DriseHm... Well I can't help you further, although I would love to, but I'm signifigantly new to RB myself. I was attempting to see if there might have been a sort of lead to go off of from line 1 and try working it from there. I'm sorry.
02:13:51RockstruckOh, that's ok!
02:13:56RockstruckI'll keep looking
02:16:37DrisePossible solution??
02:20:34krazykityou need to run to get the newer compiler
02:21:40JdGordon1S_a_i_n_t: what about this one ?
02:22:16DriseYeah, that's what the forum's solution was, and krazykit from what I can tell knows his/her stuff.
02:22:54RockstruckThanks, that seems to be the same problem I have, but I can't seem to find ~/rockbox/tools/
02:23:34Rockstruckforgot to put the ./ in front
02:23:50DriseLEt us know if it is solved.
02:24:19RockstruckWhy is it when I type a left parentheses in this web irc client that it erases my whole line?
02:24:30Drise() ??
02:24:41DriseIt works for me... What IRC do you use?
02:24:43RockstruckI'm on:
02:24:59DriseHm... I'd recommend something like XChat instead
02:26:06RockstruckWould pidgin work? I'm on windows
02:26:31DriseAs am I and XChat is free and working well for me, quite customizable too.
02:26:45DriseI'm not sure about pidgin.
02:26:51RockstruckI'll try it!
02:26:59DriseBut I get the feeling the answer would be no.
02:27:09CIA-8New commit by 03pixelma (r27544): Manual - Recording source settings description: replace target options with the correct features (needs one more in features.txt). Correct(s) wrong ...
02:28:07Rockstrucksomeone left the history in eterm still there on this vmware machine!
02:28:20Rockstrucklots of interesting dd commands
02:28:28saratogafreenode provides a webclient for their network that works better then the old rockbox one:
02:28:38CIA-8r27544 build result: All green
02:30:18pixelmaComposing sentences with correct punctuation in a list with x combinations is almost insane... and there is still one case of wrong grammar which I found too much to takje care of (The GigabeatS one will say: "The available options are: FM Radio.")
02:30:37pixelmain our manual I meant
02:31:59DriseAre you looking for help with grammar? I'm quite proficient and it is quite natural to me. Anything need a read through?
02:32:11RockstruckThanks saratoga
02:32:34DriseEarlier I was working on the Samsa page.
02:35:06 Join Metrognome [0] (48c4187a@gateway/web/freenode/ip.
02:35:42RockstruckI'm on the web client recommended by saratoga with the nickname "Metrognome"
02:35:54 Join sinthetek [0] (~sinthetek@unaffiliated/sinthetek)
02:36:04pixelmaDrise: I know it's wrong but our manuals are compiled for the different targets from one source and in the source and things will be included or excluded in a manual for a certain target depending on its capabilities. In this certain case it would just made the code even more complicated if I tried to take care of every possible thing and it's already complicated enough
02:36:38DriseAh...that has to be quite the hassle.
02:36:38 Quit Rockstruck (Quit: CGI:IRC (EOF))
02:37:44pixelmabetter than having everything seperately as you would have to keep things in sync that are the same etc. etc. :)
02:44:25MetrognomeWell, that command is still executing. Do you know how long it will take?
02:47:23 Quit DerPapst (Quit: Leaving.)
02:47:36 Quit goffa (Read error: No route to host)
02:48:06 Join mt_ [0] (~mtee@
02:50:13 Quit RandomCake (Quit: ChatZilla 0.9.86 [Firefox 3.6.7/20100713130626])
02:51:19MetrognomeI mean, the script is still running
02:53:44 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
02:59:53krazykitit'll take as long as it takes. it depends on how fast a processor you have, among other factors.
03:01:49AlexPIf you are doing it in cygwin it'll take a few years
03:04:27AlexPpixelma: I got to the point with some things of just separating whole sentences - more duplication, but next to impossible otherwise.
03:06:33pixelmaI'm sure that wouldn't have helped me here as the most complicated is taking care of all the different combinations (and having to nest \opts and \nopts)
03:07:34AlexPyeah, can be a git
03:08:13MetrognomeHallelujah, it's done!
03:09:40pixelmaAlexP: if you want to make a sentence for the GigabeatS alone (and not using "gigabeats" as option)... it's the only target that *only* has radio in the recording source settings list - neither digital, nor microphone, nor line-in
03:10:21MetrognomeThanks so much for all your help everyone!
03:11:35AlexPpixelma: Yeah, I'd just nopt it for beast (with maybe a replacement depending on the context that I haven't read), but I'm not that fussed :)
03:15:13JdGordon1rasher: is where we are now, suggestions?
03:15:19pixelmaAlexP: it's exactly the target things I wanted to get rid of (as that was one finding when checking what needs fixing for broken things if I complete the target rename). This way the list of options will still be correct for any other target in the future that only has radio there - and in case the GigabeatS gets some more of the options... even if it is wrong grammar now.
03:15:35JdGordon1would anyone complain if XML::Simple was checked for in the configure script and required to do a build?
03:15:55pixelmaAlexP: I rate correct info a bit higher though
03:15:56AlexPpixelma: It can't get any more of the options
03:16:16AlexPJdGordon1: Depends if it is a standard package on distros
03:16:33JdGordon1thats why I ask.. it apparently isnt
03:16:41AlexPThen I'd object :)
03:16:43rasherJdGordon1: well I'm still unsure what advantage using xml gives
03:16:44JdGordon1or isnt with ubnutnu anyway
03:16:56rasherAlexP: it's available in distros, just not installed by default
03:17:03 Join fdinel [0] (
03:17:04pixelmawhy not? If e.g. it has a dock connector that can be used for line-in like on some Ipods?
03:17:06AlexPrasher: That's what I meant
03:17:22JdGordon1rasher: it means that the tricky part of making the parser is done
03:17:23AlexPpixelma: Oh, I suppose. Do those exist?
03:17:32rasherJdGordon1: also, when I was doing the langfile as xml, I settled on libxml2. Can't remember why though
03:17:52pixelmaI don't know
03:18:08JdGordon1rasher: I dont want to write the script in c thoguh
03:18:16AlexPpixelma: I've never seen one (it doesn't fit the F dock), but I haven't looked hard
03:18:46rasherJdGordon1: key: value is pretty easy to parse, and much easier on the eyes :\
03:19:20AlexPI bloody hate having to edit xml by hand (just a general observation)
03:19:34JdGordon1perl paths check the pwd so maybe we just ship XML/ with the script so it gets picked up if it isnt installed
03:19:38AlexPwell, that may be a bit strong :)
03:19:38 Quit bertrik (Ping timeout: 265 seconds)
03:20:56JdGordon1rasher: how would key: value work when we want to use TeX in the value bit? need to then handlel multiline and all that
03:21:22*rasher shrugs
03:21:38pixelmaAlexP: anyway, I just wanted to get rid of target options there as they can get wrong and need fixing for future targets and there already were mistakes in there (it was confusing as you can't see why a target was in this or that list). And I just spent hours to fix it and catch every possible combination :\
03:21:51AlexPgood work :)
03:21:52gevaertsJdGordon1: you want to put TeX in xml?
03:22:07JdGordon1I dont, but others might
03:24:51 Join ForwardSpace [0] (
03:24:54 Join BHSPitLappy [0] (~BHSPitLap@unaffiliated/bhspitmonkey)
03:25:11gevaertsIsn't the manual already complicated enough without adding that sort of constructions?
03:25:49 Quit Metrognome (Quit: Page closed)
03:26:26JdGordon1trying to uncomplicate it here, I'm not saying it would be TeX but it might be useful to copy whatever the text is straight into generated TeX for use with the build
03:26:50pixelmato get only one file you need to update in case of changes from which the manual and the theme editor help and maybe the wiki can be autommatically updated
03:28:22 Join binaryhermit [0] (~binaryher@
03:28:56*JdGordon1 guesses this is getting sidelined then :/
03:30:43 Quit domonoky1 (Read error: Connection reset by peer)
03:35:44 Join Muhd [0] (ad37ecd0@gateway/web/freenode/ip.
03:36:15 Quit Muhd (Client Quit)
03:36:42 Quit saratoga (Quit: Page closed)
03:43:23JdGordon1xml - 4 lines per tag...key:value - 6+ lines per tag (average)
03:44:04 Quit Farthen (Remote host closed the connection)
03:44:07 Join Farthen [0] (
03:48:30 Join Zigtown [0] (
03:59:43 Join hebz0rl_ [0] (
04:03:04 Quit hebz0rl (Ping timeout: 276 seconds)
04:05:08***Saving seen data "./dancer.seen"
04:11:31 Quit JdGordon1 (Ping timeout: 265 seconds)
04:15:53 Quit TheSeven (Ping timeout: 260 seconds)
04:20:05 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:29:01 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:29:03 Quit amiconn (Disconnected by services)
04:29:21 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:29:51 Quit pixelma (Disconnected by services)
04:29:53 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:30:08 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:31:27 Join JdGordon1 [0] (~jonno@
04:44:53 Quit binaryhermit (Quit: Leaving)
04:53:05DriseWhile I'm here, anything new done on the Fuze V2 USB? Any updates since like a week and a half ago?
04:54:39 Join cached [0] (
04:55:42 Join Barahir [0] (
04:55:43cachedipodpatcher is crashing on me. specifically, it says "Error reading from disc: " and then the "ipodpatcher has stopped working and must be closed blah blah blah" window appears
04:59:22 Quit Barahir_ (Ping timeout: 260 seconds)
05:04:33 Quit Drise (Quit: Leaving)
05:11:10 Quit BHSPitLappy (Ping timeout: 245 seconds)
05:16:43 Quit AndyI ()
05:22:43 Quit hebz0rl_ (Quit: Ex-Chat)
05:27:19 Quit bieber (Ping timeout: 276 seconds)
05:27:32 Join bieber [0] (
05:29:43 Quit milz (Remote host closed the connection)
05:30:05maraz -n
05:30:18 Quit Zigtown (Ping timeout: 252 seconds)
05:37:28marazso, OF still works just as expected, even when additional heat is applied.
05:38:53 Join milz [0] (
05:44:26cachedhow does one exit chess >_>
05:52:51krazykitdepends on the device. the manual contains all the keymappings though, so you could easily look it up
05:54:30 Quit joecool (Quit: restarting X)
05:57:11 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
05:58:15 Quit anewuser (Quit: Sixth edition of 25 samples + 7days + Remix Phase July 19th - July 26th)
06:01:30 Quit myAbyss (Remote host closed the connection)
06:05:11***Saving seen data "./dancer.seen"
06:08:26 Join advcomp2019_ [0] (~advcomp20@unaffiliated/advcomp2019)
06:08:58 Quit ForwardSpace (Ping timeout: 265 seconds)
06:23:46 Quit mt_ (Read error: Connection reset by peer)
06:27:19maraz*PANIC* Dir entry 2 in sector 0 is not free! E1 50 4F 44
06:32:37lestatarhi gang...question re: pictureflow and gigabeatF40 please :-)
06:33:21lestatarlong press on WPS screen does not show Pictureflow as an option
06:33:36lestataron the does on the v2 Fuze...any ideas?
06:33:46 Join SurreyTrash [0] (
06:36:00SurreyTrashi noticed i dont see anything for sony walkman nwz-E344 anyone know of any similar software for that model or should i attempt to google this myself lol asking another question myself, but i own a E345...good luck...afaik, there are zero rockbox-like alternatives for sony devices
06:37:10SurreyTrashdamn that sux
06:37:30SurreyTrashsad thing is i used to own a phillips go-gear but i lost it and they have that here haha feels bad man
06:37:51SurreyTrashguess i'll have to jump on the ipod bandwagon soon an ride one of those out
06:38:06lestatarget a fuze or clip+
06:38:21lestatarrockbox works great with both! :-)
06:38:27SurreyTrashi just want somthing nice i can customize heh
06:38:35SurreyTrashorly may have to look into that then thx hehe
06:38:58lestatari got a fuze v2...with rb, it went from a very good player to an insanely great one
06:40:22SurreyTrashyeah i got bored with my mp3 player and its layout/theme and no apps i just leave it on shuffle and take it to the gym or whatever lol
06:40:32SurreyTrashtrying to get the most out if it i can but guess its time for a new one
06:41:42lestatari got 2 daps rb now, the fuze and an amazing gigabeat f40...rb truly is terrific
06:42:06SurreyTrashmust be nice lol
06:42:21lestatarit ain't bad :-)
06:42:46lestatartoo bad...don't wanna chat on here, but seems everyone is tied up to answer my question atm...
06:43:17SurreyTrashyeah thats how it goes tho
06:43:39SurreyTrashim on google looking for other alternatives someone out there has to have done somthing with sony lol worries...not life threatening...will pop in another time
06:43:54lestatar^ i would be interested to know about that
06:44:05lestatarbut i have heard zero on the sony front
06:44:44SurreyTrashwell if i ever come back an see u here i will let u know if i come across anything
06:44:59lestatargonna run....good luck with the sony search...if u do find something, post on anthingbutipod :-)
06:45:11lestatarthanks! later
06:47:47 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
06:50:15marazand that's definite skipping right there.
06:50:58marazoh well. i guess i'll have to use the OF if rockbox is b0rked
07:03:43 Join BHSPitMini [0] (
07:16:10 Quit JdGordon1 (Ping timeout: 245 seconds)
07:29:45 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
07:32:35 Quit Horschti (Ping timeout: 264 seconds)
07:40:17 Quit SurreyTrash (Quit: ChatZilla 0.9.86 [Firefox 3.5.11/20100701023340])
07:43:38 Quit bunnyboi (Quit: There are two major products that come out of Berkeley: LSD and UNIX. We don't believe this to be a coincidence.)
07:55:42 Join webguest-ph [0] (
07:56:25 Quit webguest-ph (Client Quit)
08:03:36S_a_i_n_tmaraz: Are you talking about your Nano again?
08:04:00S_a_i_n_tI *really* think the problem is in your hardware.
08:05:12***Saving seen data "./dancer.seen"
08:08:43 Quit S_a_i_n_t (Ping timeout: 252 seconds)
08:16:27 Quit CaptainKwel (Quit: Ex-Chat)
08:27:56 Quit BHSPitMini (Ping timeout: 265 seconds)
08:49:06 Join Buschel [0] (
08:49:37Buschelmaraz: just another wild guess, did you check r23740? maybe mulitsector writes are the problem for you?
08:53:43 Quit Buschel (Client Quit)
08:54:14 Join jfc [0] (
09:19:12 Join S_a_i_n_t [0] (S_a_i_n_t@
09:35:31 Join stoffel [0] (
09:36:33bieberbluebrother: QuaZip is working great for output as well. Looks like it does zipping and unzipping correctly, and it works as a QIODevice to boot :)
09:39:48 Quit bug2000 (Ping timeout: 248 seconds)
09:50:42 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
09:58:28 Join Rob2222 [0] (
10:01:28 Quit Rob2223 (Ping timeout: 265 seconds)
10:05:16***Saving seen data "./dancer.seen"
10:21:05 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
10:24:25 Quit BlakeJohnson862 (Ping timeout: 265 seconds)
10:25:46marazi would check r23740, but as i already installed the newer toolchain (r23740 still uses the older, testing arm-eabi toolchain), it didn't want to compile
10:26:45TheSevenjust compile it using the old toolchain then
10:26:56TheSeven(the non-eabi one)
10:27:07marazi tried, for some reason it did not work either
10:27:14*TheSeven wonders what broke it
10:27:31TheSevenbtw, what's your nand id? is it a parallel or cached write chip?
10:27:41marazno idea. how do i find out?
10:28:17TheSeveniirc it's somewhere in debug=>show hw info
10:29:13marazis r23740 before or after 3.6?
10:29:27S_a_i_n_tAFAIK only the Nano2g gas that in the debug
10:29:43marazthing is, both disk mode and original firmware work perfectly
10:29:54 Join bug2000 [0] (~bug@unaffiliated/bug2000)
10:29:59 Join ender` [0] (
10:30:02marazso it's not likely that the nand has worn out yet
10:30:29S_a_i_n_tThe thing is though, that RB and Apple may be doing things in different ways.
10:30:30TheSeventhere might be some corruption though
10:31:07marazyeah, but i mean, there is absolutely *nothing* wrong with them, but the instant i boot rockbox and write something to the disk...
10:31:17TheSevenwhich target are we talking about? nano 1g or 2g?
10:31:54marazwonder if there's a patch to disable multisector writes
10:32:02TheSevenah, forget my comments about the nand id then
10:32:05marazor even better, a compiler flag? :3
10:33:21TheSevendid you try just reverse merging that revision?
10:33:35marazhuh? :)
10:33:46marazi don't quite feel at home with svn yet...
10:34:02S_a_i_n_tthen use git ;)
10:36:11 Join JdGordon1 [0] (~jonno@
10:36:22marazyeah, i'm trying to configure r23739 and it tells me that the compilter (arm-elf-gcc) is not in my path...
10:36:55S_a_i_n_tadd it to the path then?
10:38:27S_a_i_n_tI can't remember where the old toolchains installed to, otherwise I'd just say.
10:39:41maraz/rockbox-23739/apps/action.c:26:18: error: lang.h: No such file or directory
10:40:53S_a_i_n_tare you doing the "make" in a new directory? if not, you probably need a "make clean" or, "make veryclean"
10:43:13 Quit JdGordon (Disconnected by services)
10:43:19 Nick JdGordon1 is now known as JdGordon (~jonno@
10:44:03marazah, now it's compiling
10:44:16 Join JdGordon1 [0] (
10:44:19marazshould i compile the bootloader too?
10:45:29S_a_i_n_tThere's no need, as long as you're using a recent one I wouldn;t bother.
10:46:17S_a_i_n_tThe only realy need for compiling your own bootload is a: if you've changed the operation of it in some way, b: if you're using an advanced OSOS install
10:46:37S_a_i_n_t*bootloader (not my day for spelling apparently)
10:47:56 Quit mc2739 (Ping timeout: 240 seconds)
10:49:46*maraz installs
10:50:10 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
10:50:56S_a_i_n_tmaraz: It may not hurt to do a complete, clean, install either.
10:51:09marazwhich is what i'm doing.
10:51:09S_a_i_n_tAs in, abck up what you need to, and delete the .rockbox dir.
10:51:14S_a_i_n_tAh, awesome.
10:52:07S_a_i_n_tIt's funny, it really shouldn't make a difference but I have seen a simple clean install fix up al sorts of weird problems before.
10:56:24marazso far, so good. listening for corruption now
10:56:41marazbut at least it didn't boot up with "no .rockbox directory" or corrupted fonts or anything
10:57:44S_a_i_n_tI realise that your unit seems to run the OF fine, but, you must admit that it seems a little odd that so many other users don't experience this too right?
10:57:52S_a_i_n_tthat's what makes me want to look at hardware
10:57:58marazit is indeed peculiar
10:58:05S_a_i_n_tFor some reason your unit can't handle something many others can.
10:58:08 Quit Strife89 (Read error: Connection reset by peer)
10:58:15marazhowever, i have run other builds of rockbox just fine
10:58:34marazthere could be a minute hardware difference
10:58:42marazmy unit *is* one of the very earliest shipped out
10:59:26S_a_i_n_tfrom what I know the Nano1G never changed much, or at all. The only difference I have personally seen is two slightly different batteries.
10:59:49S_a_i_n_t(which, are both basically the same...just, different)
11:01:40 Join shai [0] (
11:08:13*maraz tries heating the unit
11:10:12marazthe obvious next step would be to try r23740 and see if it misbehaves
11:10:49 Quit mc2739 (Ping timeout: 240 seconds)
11:11:44 Join BlakeJohnson86 [0] (
11:12:36marazwith r23739 and with the back cover hot to the touch... no corruption whatsoever
11:12:53maraz(which is nice, since if nothing else works, i can stick with this ;)
11:13:01 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
11:14:50 Join slck [0] (Venci@Slackware.SlackPix.Com)
11:19:56 Quit GodEater (Read error: Operation timed out)
11:21:23 Join GodEater [0] (
11:21:23 Quit GodEater (Changing host)
11:21:23 Join GodEater [0] (~bibble@rockbox/staff/GodEater)
11:26:05marazstrangely, r23740 seems to be working too.
11:29:45*TheSeven suggests brute force bisecting then
11:30:25marazso, jump up n revisions until problems start?
11:30:35marazor down from current until they stop?
11:31:05TheSevendivide'n'conquer will be fastest if you don't have a suspicion which revision could be the cause
11:31:20marazalgorithms 101
11:31:37TheSevenso always check the one in the middle between the last known-working and first known-failing one, until it's down to a single revision
11:52:27maraz25642 is broken
11:53:24maraz23691 next, then
11:53:55marazwait, 24691
11:54:29bertrikare you going to mention each version you try here ... ?
11:55:59marazsorry, thought someone was interested
11:56:02marazwill shut up now :)
12:05:19***Saving seen data "./dancer.seen"
12:17:02 Quit BlakeJohnson86 (Ping timeout: 265 seconds)
12:27:09 Join BlakeJohnson86 [0] (
12:28:24 Join Giova [0] (~giovanni@
12:43:54 Quit S_a_i_n_t (Ping timeout: 246 seconds)
12:44:00 Join S_a_i_n_t [0] (S_a_i_n_t@
12:45:09 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:08:35 Join Kitr88 [0] (
13:10:56 Join DerPapst [0] (
13:11:43 Quit Kitar|st (Ping timeout: 260 seconds)
13:13:23 Quit Kitr88 (Ping timeout: 276 seconds)
13:15:49 Quit bluebrother (Ping timeout: 240 seconds)
13:17:57 Join Kitar|st [0] (
13:25:44 Quit balintx (Ping timeout: 260 seconds)
13:30:24 Join balintx [0] (
13:30:57 Join Jerom [0] (~jerome@
13:34:08JdGordonhow terrible would it be if viewports which are no longer enabled get cleared?
13:34:19JdGordonAssuming people set up viewports correctly this should never be a problem, right?
13:39:27 Quit stoffel (Ping timeout: 265 seconds)
13:39:51JdGordonanyone seen this before...
13:39:53JdGordonrebuffer_handle(): space is needed
13:39:54JdGordonbuffer_handle(): Preventing handle corruption: copy_n:32768 overlap:28020 h1.filerem:719224
13:43:42Giovahd JdGordon
13:43:53GiovaI've got a question
13:44:16JdGordonlucky you
13:44:34Giovathe onda vx747 and 777 has got an external speaker, could it be possible to add to the wps structure a tag to activate and deactivate it?
13:45:17JdGordonyes, if someone can be bothered doing it
13:46:10GiovaI will try to look at the code, but I'm not really god at coding, so I will need some advice
14:02:29 Join ThomasAH [0] (
14:02:54 Quit ThomasAH (Client Quit)
14:05:23***Saving seen data "./dancer.seen"
14:06:51 Join ThomasAH [0] (
14:22:56 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
14:28:16 Join funman [0] (~fun@rockbox/developer/funman)
14:34:59S_a_i_n_tJdGordon's busy trying to slot a new parser into the existing skin system, at the present point I'm not sure adding any new tags is very high on the list.
14:35:05S_a_i_n_tGiova: ^
14:36:07S_a_i_n_tre: <JdGordon> how terrible would it be if viewports which are no longer enabled get cleared?
14:36:13S_a_i_n_tSeems a solid idea to me.
14:36:31JdGordonI'm hoping that only breaks people being silly
14:36:35S_a_i_n_tIf used correctly, then, it should never be a problem.
14:36:47JdGordonbut there is a tiny chance someone is relying on that behaviour to do interesting effects
14:37:06S_a_i_n_tI know I have before...and probably do still.
14:37:15S_a_i_n_tBut, I live to find holes in the skins ;)
14:37:25S_a_i_n_tI'll find a new one :D
14:38:45S_a_i_n_tActually, I can think of one theme of mine that would break, but, sublines in conditionals *should* allow me to get around it now...if not, it's an old theme anyway.
14:38:57 Join einhirn [0] (
14:41:01 Join einhirn_ [0] (
14:41:12S_a_i_n_tmaraz: You are aware *what* makes the heat one the back plate of a Nano1G right?
14:41:20S_a_i_n_tIf you're's the battery.
14:43:17 Quit einhirn (Ping timeout: 245 seconds)
14:43:47S_a_i_n_tThere is essentially nothing between the backplate and the battery, during charging and high-draw the battery can put out a fair ammount of heat.
14:47:00kugelJdGordon: yes, I've seen that before
14:47:50kugelwhat you wrote about the buffering debug output
14:48:05JdGordonah, any ideas what is was about? other than a possible bug?
14:48:49kugelit's pretty discriptive, isn't it? :)
14:50:04 Join fdinel [0] (
14:50:15kugelit's been some time, but IIRC that happens when buffering is out of space. it tries to buffer the next handle (only the struct) nevertheless, which may overwrite the previous handle
14:51:17JdGordonsomething is breaking my pointers :<
14:53:35gevaertsJdGordon: for the record, I think that breaking "features" that (a) were not intended originally and (b) can be considered as bugs (I still think not clearing viewports when disabling them is a bug, not a feature :) should be perfectly OK
14:54:22*S_a_i_n_t would prefer a tag to specifically clear them though, to keep both camps happy
14:54:25JdGordonyeah, I agree, but I did have the goal of not requiring anything to change on the users side
14:54:30JdGordonwhich is obviously not going to happen anymore
14:55:27*JdGordon is currently fighting with the list title text tag :/
14:56:00JdGordoncan I get gdb to break if anything writes to a memory address?
14:56:57gevaertsJdGordon: make the svn parser/renderer clear the viewports. Then the new one won't change this :)
14:57:20JdGordoneasier said than done :)
14:57:32JdGordonand I'm almost at the breaking point with this bloody patch!
14:57:38JdGordonif it isnt done by mid week I might just go crazy
15:00:03 Quit fdinel (Read error: Connection reset by peer)
15:03:41 Quit einhirn_ (Read error: Connection reset by peer)
15:06:44 Quit JdGordon1 (Quit: Leaving.)
15:12:17 Join anewuser [0] (anewuser@unaffiliated/anewuser)
15:15:09 Quit cached (Read error: Connection reset by peer)
15:20:40 Quit Jerom (Quit: Leaving.)
15:21:30 Join Jerom [0] (~heidi@
15:27:16 Join mitk [0] (
15:28:39mitkJdGordon: ping
15:29:16mitkCan you compile 11470 for clip+ and Fuze v1? I've got compilation errors
15:30:07JdGordonok hang on
15:31:42JdGordonah bugger... sync messed u
15:34:19JdGordonmitk: reuploaded the patch so it compiles
15:34:42mitkWill try then. cu">
15:38:19JdGordonfuze coming up also">
15:40:28kisakhas anyone made a bug entry simular to wine bug 10000?
15:41:01 Join pamaury [0] (
15:41:01 Quit pamaury (Changing host)
15:41:01 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
15:41:53JdGordonprobably.... although I have no idea what that bug is
15:42:22kisak" Bug 10000 - The original win32 api implementation is still more popular than wine."
15:42:48JdGordonok, then why would anyone create a bug like that on rockbox?
15:43:13*JdGordon finds a potentially nice optimisation....
15:43:26JdGordon~300 params are alloced in each of my example skins
15:43:52kisakbasicly as a stub to bounce ideas around on how to increase the spread of rockbox
15:43:54JdGordonassuming they are never used again that is a 3900byte waste
15:44:09S_a_i_n_tkisak: The tracker is for actual bugs
15:44:23S_a_i_n_tthat would not be a bug.
15:44:25JdGordonspreading rockbox means more users....
15:44:33CIA-8New commit by 03kugel (r27545): $rbdir => $temp_dir where appropriate, shadowing $rbdir with the temp dir broke (for rbdir != ".rockbox")
15:44:36JdGordonit is very debateable that that is a good thing
15:44:48S_a_i_n_tMore users, more bugs :P
15:45:08 Quit GodEater (Read error: Operation timed out)
15:45:23funmankugel: did you have plans to remove
15:45:25mitkJdGordon: Compiled my own and got your builds. Hard point is to make decision which one use to tests. I'm going to think about it :)
15:45:33kugelfunman: not recently, no :)
15:45:56maraz okay
15:46:10marazanyone who told me this was a hardware problem, nuts :)
15:46:13CIA-8r27545 build result: All green
15:46:42marazr24404 is stable as a rock even if i heat the back of my nano hot to the touch
15:46:51S_a_i_n_tmaraz: Riddle me this then, whay are my 4 Nano1Gs fine then?
15:46:52marazr24405 is extremely unstable and corrupts
15:47:08marazcoincidently, r24405 introduces some ATA DMA changes on pp5020
15:48:32bertrikmaraz, good to hear you pinpointed it :)
15:48:57marazi wouldn't exactly call it pinpointing >_>
15:49:20JdGordontoo easy :) 46.7KB -> 42.9KB skin usage
15:49:20marazbut at least it is not confirmed that it's not a hardware problem
15:49:22JdGordonstill double svn though
15:49:52*S_a_i_n_t is not so sure, I would have expected to have seen some tiny hint of this in any of my Nano1Gs, but, I haven't
15:50:11marazit is baffling
15:50:13S_a_i_n_tthere has to be a reason why your Nano is not performing, and others are.
15:50:19marazi agree
15:50:21funmanmaraz: could be a hardware problem specific to DMA
15:50:23 Join dockimble [0] (~dockimble@
15:50:24S_a_i_n_tIt's running the same code, so...only hardware is left.
15:50:58marazi think i'm going to make a current build without ata and try it
15:51:02marazer, without ata dma :)
15:51:18S_a_i_n_tApparently RB is doing something differently than the OF, which the other Nanos can handle...but for some reason yours can't.
15:51:24S_a_i_n_tIt'd be nice to know what that is.
15:51:45marazi'm all for finding out.
15:52:05S_a_i_n_tWell, jsyk...I nuked a Nano trying ;)
15:52:14S_a_i_n_tDon;t go *that* far :P
15:52:43S_a_i_n_tThe only thing I can say is that excess heat didn't seem to be a factor.
15:52:50maraznot in your case, no
15:52:57marazin my case heat greatly accentuates the problem
15:53:27S_a_i_n_tMy clickwheel ended up oval, and I couldn't touch the case without a towel and still maganed playback and transfer.
15:53:44S_a_i_n_t(this was at 50celcius)
15:53:59S_a_i_n_tbah, *managed
15:54:15CIA-8New commit by 03funman (r27546): Remove rockbox 2.6 plugin API doc
15:55:05bertrikthe DMA timing depends on the DMA mode, and this DMA mode depends on the disk used
15:55:12S_a_i_n_tThe only thing I can think of that could make the heat on the back panel (look at a layout of the mainboard), is the battery...perhaps you have issues powering the device?
15:55:39maraznot that i know of
15:55:40S_a_i_n_tMaybe revision a needs more juice then revision b?
15:55:48S_a_i_n_t(pure guesswork here...)
15:55:51CIA-8r27546 build result: All green
15:56:15S_a_i_n_tthe battery *is* 5 years old...
15:56:16marazthen again, that problem would have been fixed by the underclocking, yes?
15:56:27S_a_i_n_tNot necessarily.
15:56:35marazrockbox reports ~9h life for 100% battery, though i'm not sure how accurate that is
15:58:21bertrikthe 9h is not calculated from the actual battery curve in real-time or something like that
15:59:14S_a_i_n_tmaraz: jsyk I'm getting 523 minutes at 82%
15:59:33marazbeen looking around for a new battery
15:59:35S_a_i_n_tso, your battery is definitely lesser than the one I have in front of me now.
15:59:44S_a_i_n_tIf that matters or sot, whio knows.
15:59:49 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
16:00:06marazit is worth a try
16:01:17S_a_i_n_tAnother Nano1G reports 503 min at 77%, still *slightly* better, but, not as much as the first.
16:02:09S_a_i_n_tThose are the only two I have with standard batteries though so I can't give any more (relevant) results.
16:02:24S_a_i_n_tThe other two have Nano2G batteries in.
16:03:17JdGordonbieber: any ideas how to get these figures down...? 629 bytes alloced for strings
16:03:17JdGordon20736 bytes for 576 elements
16:03:20gevaertsmaraz: maybe you could experiment a bit with firmware/target/arm/ata-target.h
16:03:44bertrikIs there a way in rockbox to tell which mode (DMAor PIO and what speed) and the ATA interface is in?
16:03:46gevaertsTry setting ATA_MAX_UDMA to 1
16:04:14*JdGordon guesses not possible
16:05:26***Saving seen data "./dancer.seen"
16:06:27marazhm, so UDMA 2 is stable at 30MHz and 1 at 24?
16:06:49 Join wodz [0] (
16:07:08gevaertsTests have shown that 2 is stable at 30MHz, but possibly that's not true on your nano
16:07:45bertrikRuntime runtime estimation would be nice
16:08:21S_a_i_n_tbertrik: FS #10890
16:08:31 Quit JdGordon (Quit: Leaving.)
16:08:34S_a_i_n_t(dynamic runtime estimation)
16:08:37wodzCurrent codec optimizations on coldfires gives strong argument to lower normal frequency to 34MHz (at least on greysacale targets). Most of the common formats will still decode unboosted
16:08:42bertrikI think we would nice some kind of robust regression for that
16:09:32S_a_i_n_tfs#10890 is *surprisingly* accurate on my Nano2Gs
16:11:10bertrikS_a_i_n_t, that doesn't work on a lot of targets, because it needs the ability of the target to read its own battery current
16:11:36bertrikand many can't do that
16:11:45S_a_i_n_tyeah, good for the ones that can the ones that cannot could do it is beyond me.
16:13:06bertrikS_a_i_n_t, we can monitor voltage which is a semi-OK indicator for battery charge left for lithium batteries (we already calculate percentage from battery voltage)
16:13:32bertrikSo if we have a robust estimate of the rate of change of capacity, we can extrapolate to the point in time where the charge will be 0
16:14:07wodzisn't it how it is working now?
16:14:44S_a_i_n_tmaraz: You can force the CPU frequency to 24Mhz on the Nano1G without changing any code, but, you need to do it each boot as I understand it.
16:15:08bertrikno, each target has a hard-coded capacity and a hard-coded "normal" current, given a hard-coded "normal" runtime which is multiplied by the charge percentage
16:15:14S_a_i_n_tJust go to the dubug/cpu frequency, make sure it is on 30Mhz, and press select.
16:15:19gevaertsS_a_i_n_t: that's not really the issue though
16:15:22S_a_i_n_tThis will drop it to 24Mhz
16:15:46S_a_i_n_tgevaerts: But, at 24Mhz it will use UDMA 1, no?
16:15:51funmani want to move some GCC features to a specific header, any hint for the name? i was thinking of compiler.h but as it might already exist, perhaps gcc_extensions.h
16:16:12wodzgcc_compiler.h ?
16:16:23gevaertsIt uses UDMA 1 if CPUFREQ_NORMAL is 24MHz, not if you force the frequency down
16:16:38S_a_i_n_tAaargh, yep...quite correct.
16:17:06funmanwodz: ideally it could also support other compilers and i think that name imply it's only needed for gcc
16:17:46wodzfunman: what else free compiler is possible to use?
16:18:14funmanfree compiler: Code Sourcery GCC (which is also gcc), but we could also use non free compilers (for own use)
16:19:22wodzwhat's the point of using non free compiler other than to look at compiler specific quirks?
16:19:49wodznot for release
16:19:59funman< funman> .. for own use
16:20:19funmananyway a lot of work is needed for working with other compilers
16:20:29funmanfor now i just want to try code sourcery gcc
16:20:36wodzthats my point
16:20:50funmanand i need to move some definitions anyway
16:21:28funmanbecause their compiler ships _ansi.h, so i need to move ATTRIBUTE_(PRINTF|SCANF) which doesn't belong here
16:22:01 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
16:22:38wodzI would say we use gcc - if someone want's to play with other compiler go ahead but without messing with code
16:22:55wodz(our code)
16:23:10funmanit's my code too
16:23:36wodzsupporting different gcc versions is enough IMO
16:24:33funmani don't know which compilers previous rockbox hackers used but ATTRIBUTE_PRINTF seems to imply some didn't use gcc
16:25:00wodzmy opinion is that unless we have real alternative to gcc we should leave things as is
16:25:38funmani'll make the change to support code sourcery gcc
16:29:09CIA-8New commit by 03kugel (r27547): Use MAX_PATH for array used for filename.
16:30:33wodzis there a way to do sound setting in log scale (not linear)?
16:30:46CIA-8r27547 build result: All green
16:32:06wodzaudiohw_settings seem to accept only linear settings
16:32:41marazgevaerts: forcing ATA_MAX_UDMA to 1 doesn't help, but leaving #define HAVE_ATA_DMA away from /firmware/export/config/ipodnano1g.h does. i'm assuming it falls back to PIO transfers?
16:32:51gevaertsit does, yes
16:33:55marazof course it's possible that the aging NAND is no longer keeping in spec and it just isn't apparent at PIO
16:34:37marazoh well
16:34:44marazat least i have a working player now
16:34:52*maraz knocks on wood
16:34:59S_a_i_n_twith 10 minutes testing ;)
16:35:04marazheh, yeah
16:35:21marazthen again, it's pretty apparent right off the bat if there's something wonky going on
16:35:39marazi.e. theme doesn't load, corrupted fonts, etc
16:37:20marazstill, it's likely there are more cases, they just haven't popped up - maybe the owners thought rockbox is normally extremely buggy ;)
16:38:24bertrikmaraz, my guess is that changing ATA/PIO timings fixes the problem, then it's some kind of bus timing issue and not related to aging NAND
16:38:55marazyeah, as the problems only exist in rockbox, i have to agree
16:40:10bertrikBut I'm not familiar with the nano 1g, I'm not sure even why it would use the ATA bus when it has NAND memory for storage
16:41:09marazprobably for compatibility with other apple hardware
16:41:34marazin other words, similar firmware for all ipods?
16:44:05 Join panni_ [0] (
16:44:32CIA-8New commit by 03funman (r27548): Move some gcc extensions to new gcc_extensions.h header ...
16:45:01 Join Buschel [0] (
16:45:10 Quit kugel (Remote host closed the connection)
16:45:22 Join kugel [0] (~kugel@rockbox/developer/kugel)
16:46:15Buschelmaraz: good to see that you found a way to solve your issue
16:46:36CIA-8r27548 build result: 518 errors, 10 warnings (funman committed)
16:50:29 Quit wodz (Quit: Leaving)
16:51:10bertrikI think I'll work on a patch for "backlight fading" for the clips
16:52:02bertrikand create a nice exponential brightness curve at the same time
16:53:08CIA-8New commit by 03funman (r27549): strncpy.c / assert.h : need to include _ansi.h
16:55:13CIA-8r27549 build result: 116 errors, 10 warnings (funman committed)
16:55:21 Join MethoS- [0] (~clemens@
16:55:22funman <- diff between GCC & code sourcery GCC (doesn't go past logo on Clip+)
16:56:27 Join domonoky1 [0] (
16:57:16kugelfunman: I had CS GCC working on my fuze once
16:58:04funmandid it require much hacking in the source?
16:58:50 Quit domonoky (Ping timeout: 265 seconds)
16:59:18kugelit just has the unwinding stuff in, so it needs some linker script changes (and a abort() stub)
16:59:38kugelor had, I used the 2009q3 version
17:00:07kugelI'm suprised they didn't update the gcc & binutils version in 2010q1
17:03:10 Join CaptainKwel [0] (
17:03:25funmani can't find where rockbox adds -I$(FIRMDIR)/include
17:04:03funmanah firmware.make
17:06:58 Quit panni_ (Read error: Connection reset by peer)
17:07:24CIA-8New commit by 03funman (r27550): include firmware.make unconditionally ...
17:08:28 Join panni_ [0] (
17:09:22CIA-8r27550 build result: 128 errors, 10 warnings (funman committed)
17:10:00funmanwhy do i always test the target which works by incidence? :o)
17:11:58funmanerr, also why did the player build work when i tested it but not anymore :o
17:13:23funmanh10 checkwps build succeeds despite the #error
17:13:54funmanah, it's only when generating dependancies
17:22:49CIA-8New commit by 03funman (r27551): Revert r27550 : we don't want to generate dependancies for firmware/
17:22:53CIA-8New commit by 03funman (r27552): database: fix include dirs
17:24:49CIA-8r27551 build result: 116 errors, 10 warnings (funman committed)
17:26:22kugelfunman: having a nice time? :D
17:26:43*funman explores the depths of rockbox make system
17:26:43CIA-8r27552 build result: All green
17:27:13funmanbtw i didn't understand the -fno-builtin of checkwps so i just left it there
17:28:25CIA-8New commit by 03funman (r27553): bootchart.h doesn't need _ansi.h, especially not in a relative path
17:30:11CIA-8r27553 build result: All green
17:32:07 Quit DerPapst (Quit: Leaving.)
17:40:16kugeloh damn, the themes from the theme site don't work in the app :\
17:50:06 Join VoivoD [0] (
17:55:30bluebrotherbieber: are you aware of changes made to the findreplacedialog implementation between the "old" one and the currently imported one? I can't build it on Windows anymore :(
17:59:05kugeldo we use gcc extentions? like ?:-operator extension?
18:00:20funman__attribute__(()) for a start
18:01:41kugelfunman: __attribute__ is more portable though, other extensions are not (e.g. binary numbers or the ?: one)
18:02:59funmanno '?:' according to grep
18:04:54bluebrothermy K&R lists :? in the Conditional Expressions sections so it can't be a gcc extension
18:05:29bluebrotherand we can't use binary prefixes because the gcc (still) used for Coldfire is too old to support them.
18:05:31***Saving seen data "./dancer.seen"
18:07:58bluebrother(just checked: binary literal prefixes were introduced with gcc 4.3)
18:12:13kugel?: itself is not an extension, but gcc has an extension for it
18:12:50kugelwe could use binary numbers in the target tree. where they are most useful anyway (for hw register bits)
18:13:53kugelbluebrother: you can do A = B ?: C, which is equivalent to A = B ? B : C
18:16:01bluebrotherkugel: oh, I though you were talking about a = b ? c : d. That extension seems a bit useless to me.
18:16:29bluebrotherbinary literals ... well, at least I prefer shifting if I really want to see the bits.
18:17:53kugelif B is very long, the extension is handy
18:19:19bluebrotherif the expression is very long it might be a better idea to write it as if-else anyway. IMHO :)
18:20:38gevaertsI'd say the gains don't warrant the extra syntax to learn
18:21:12 Join geekhead [0] (
18:21:22bertrikthe actual brightness vs. contrast setting curve for the clip+ is far from linear
18:22:43 Quit geekhead (Client Quit)
18:23:18bertrikI tried to determine the brightness by taking photos and measure it with gimp, but I realise now that JPG probably also does some non-linear thing to brightness
18:27:43CIA-8New commit by 03kugel (r27554): Rewrite r21464 in a more compact/readable way. Reduces code duplication as well.
18:29:51CIA-8r27554 build result: All green
18:40:41CIA-8New commit by 03kugel (r27555): A bit more cleanup in rockbox_browse().
18:42:23CIA-8r27555 build result: 0 errors, 14 warnings (kugel committed)
18:43:31GiovaThere's a way to update a theme in the theme section of the site?
18:44:05kugelthat's a false positive, gcc!
18:47:50CIA-8New commit by 03kugel (r27556): Add default case back to silence gcc false positive abount ext and dir being used uninitialized.
18:49:29CIA-8r27556 build result: All green
18:51:54amiconnbluebrother: the extension isn't useless. it avoids evaluating 'B' twice (which would duplicate possible side effects)
18:54:28 Quit pamaury (Remote host closed the connection)
18:57:37CIA-8New commit by 03kugel (r27557): Move the array to where it belongs
18:59:10CIA-8r27557 build result: All green
19:03:35 Quit Jerom (Read error: Connection reset by peer)
19:04:20 Join Jerom [0] (~jerome@
19:07:58CIA-8New commit by 03kugel (r27558): Rockbox as an application: Fix a few yellows
19:09:39CIA-8r27558 build result: All green
19:17:38 Join hebz0rl [0] (
19:19:19Giovamik: thanks
19:22:44 Quit bieber (Ping timeout: 260 seconds)
19:23:01 Join bieber [0] (
19:27:30 Join stooo [0] (
19:31:04 Part stooo
19:32:40 Join BHSPitMini [0] (
19:34:28 Quit Buschel (Ping timeout: 265 seconds)
19:45:50 Join Jaykay [0] (
19:47:06 Join n1s [0] (~n1s@rockbox/developer/n1s)
20:04:42 Quit z35 (Ping timeout: 265 seconds)
20:05:33***Saving seen data "./dancer.seen"
20:06:43CIA-8New commit by 03bluebrother (r27559): Theme Editor: fix building on Windows. ...
20:08:22CIA-8r27559 build result: All green
20:11:12 Join Buschel [0] (
20:15:12mitkThere is no information about updating themes on themes site. Or I can't find it. Maybe putting link content from 19:18 today on Rockbox Themes will avoid many questions from theme makers?
20:15:59S_a_i_n_tWhat questions do you have?
20:16:20S_a_i_n_tAFAIK very little, if anyone, has problems uploading themes
20:16:36mitkI answered the question from Giova, today 18:43
20:19:33S_a_i_n_tThe only thing I can think of that is slightly difficult (or could be) is resubmitting themes.
20:19:33S_a_i_n_tAnd, as far as I rememebr that is answered in the forums.
20:19:33GiovaTake a look at my themes in onda vx777 section
20:19:33S_a_i_n_tAny particular reason?
20:19:33mitkIn the forums yes, but not on themes site itself
20:19:34DBUGEnqueued KICK S_a_i_n_t
20:19:34S_a_i_n_tI don't necessarily think it should be on the themesite...but, "meh"
20:19:38GiovaI've made a lot of improvements in the fulltouch theme, but as now, I will not upload it since it is the final version, if I could just update the same one, I will upload weekly
20:19:58S_a_i_n_tThehesite covers uploading themes, good searching, or searching at all covers resubmitting them.
20:20:20S_a_i_n_tGiova: And, you *can* update the same one.
20:20:45Giovaok, sorry
20:20:56GiovaI will surely find the way to do it
20:21:04mitkS_a_i_n_t: I'm talking about updating, not uploading themes
20:21:14S_a_i_n_tUpload it with the same name, and username, email address...and it will *replace* the old theme.
20:21:32mitkGiova: Take a link I given you, please
20:22:13Giovayes yes, now I've got no big time to work on this, but thanks, I will for sure
20:22:22 Quit seani (Ping timeout: 258 seconds)
20:24:52 Quit mitk (Quit: Leaving)
20:26:13 Join T44 [0] (
20:28:44CIA-8New commit by 03kugel (r27560): Use browse_folder() for browsing eq presets and languages.
20:30:03 Quit Topy (Ping timeout: 265 seconds)
20:30:24CIA-8r27560 build result: All green
20:31:55kugelgevaerts: I have converted almost all paths in the core now
20:32:08 Quit Giova (Quit: Sto andando via)
20:32:33kugelfm presets and recording dir paths are left, but RaaA doesn't have those so I'm not sure if it makes sense converting them now
20:32:44kugeland all plugins (RaaA doesn't have plugins yet)
20:33:05 Join stripwax [0] (
20:34:06CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
20:34:06*gevaerts still thinks that most plugins don't make much sense on RaaA anyway
20:34:18kugelI agree
20:34:41bluebrotherdon't you want to have an environment where you can finally run doom again? ;-)
20:35:05 Join Strife89 [0] (
20:35:08gevaertsbluebrother: sure, but all potential targets for RaaA have a native doom port already :)
20:35:23 Join seani [0] (~seani@
20:35:56bluebrothergevaerts: bah, who wants native ports? :D
20:36:55 Quit Buschel (Ping timeout: 248 seconds)
20:36:56n1ssome plugins could be fun, but most would be pointless
20:37:16n1stest_codec could be nice too
20:39:36*bluebrother grumbles at setupapi :(
20:40:41 Quit BlakeJohnson86 (Ping timeout: 265 seconds)
20:41:54funmanmost plugins aren't pointless already on current targets? O:-)
20:42:25S_a_i_n_tOf course not, think of quality like rockblox1D
20:43:07bluebrotheror pictureflow!
20:43:07n1sfft for example would make sense, but games, etc maybe not
20:43:09*bluebrother hides
20:43:37*S_a_i_n_t slaps bluebrother with a flowing picture
20:51:18kugelgevaerts: I think I'll write some kind of bootloader for android, that should help getting critical stuff like lcd to work early
20:51:31 Quit stripwax (Quit:
20:53:40*kugel still needs to work out how to make apks without eclipse
20:54:51 Join bilditup1 [0] (
20:55:25 Join BlakeJohnson86 [0] (
20:55:51 Join B4gder [0] (~daniel@rockbox/developer/bagder)
20:57:26bilditup1i have been trying to run to install the eabi toolchain. it quits, saying that '"patch" is required for this script to work.'
20:57:42S_a_i_n_tbilditup1: There's 134 people online here, if you have a question...go for it.
20:57:50S_a_i_n_tAh, a little late...sorry.
20:57:55bertrikbilditup1, that's probably correct
20:58:12 Join arcnaut [0] (~clementsp@
20:58:15bilditup1except i don't know what 'patch' is. a google and forum search has led me in circles a bit
20:58:32S_a_i_n_tthe GCC patch?
20:58:33B4gderpatch is a tool
20:58:34bilditup1(obvi have not compiled in over a month)
20:58:56S_a_i_n_tYes, ...yes it is.
20:59:28kugelbilditup1: apt-get install build-essential if you're on a debian (or something based on it) box
21:00:15bilditup1oh no, this is cygwin. sorry for skipping that. no linux/osx boxes around atm
21:00:36bluebrothercygwin has a package called patch.
21:00:52funmanyou'd spend less time installing a virtual machine running ubuntu than compiling the cross compilers on cygwin
21:00:55mc2739after updating my dev system to Ubuntu 10.04, I get this: when cross-compiling win32 sims. Any ideas on a fix?
21:01:01B4gderexpect to take a looooong time on cygwin
21:01:10bilditup1yeah, i saw, several hrs
21:01:19bilditup1i don't have the disk space for a virtualbox install
21:02:02funmanmc2739: rename midiutil.c printf to something else
21:02:02bluebrotheryou're sure you have enough space to build the compilers on cygwin then (or anything else that isn't really small)?
21:02:31S_a_i_n_tbilditup1: How much free space do you have?
21:02:32funmanlike 'midi_printf' if you can't think of something better
21:02:42 Quit bilditup1 (Read error: Connection reset by peer)
21:03:11 Join bilditup1 [0] (
21:03:27bilditup1sry. client died for a second. i'm gonna try it...
21:04:10S_a_i_n_tbilditup1: Is the package list on the cygwin wiki incomplete?
21:04:25*S_a_i_n_t is pretty sure it has devel - patch listed
21:05:38funmanmc2739: IMO the function should have "debug" in its name
21:05:47bilditup1it says 'patchutils'. sorry for not assuming that that's the same thing
21:06:07S_a_i_n_tit isn't
21:06:10bilditup1and i'm down to 20gb. need to offload some frivolous tv/movies. would rather not put a virtual machine on that last 20gb
21:06:21S_a_i_n_tbut, the dependencies should have been automatically included
21:06:28funmanbilditup1: the machine should fit in 5GB
21:06:54S_a_i_n_tA cygwin install is less than 1GB, including the source code for rockbox :P
21:07:00S_a_i_n_t(if space is an isue)
21:08:16funmanwho maintains the vmware(?) image used for rockbox dev ?
21:08:22bilditup1hmm. i've already got cygwin, and was using it for compiling from apr-june, so i didn't look at the cygwin install guide again till recently. i may still go the ubuntu route, i spose.
21:08:36S_a_i_n_tfunman: noone afaik
21:09:32bilditup1this was the page i was looking at, btw.
21:09:47S_a_i_n_tbilditup1: Needing "patch" installed isn;t a recent thing for CygWin...if, that's what you were saying.
21:09:53S_a_i_n_t(I didn't parse that too well)
21:09:54 Quit bilditup1 (Read error: Connection reset by peer)
21:10:22 Join bilditup1 [0] (
21:10:28bilditup1dang it, i need a better client.
21:10:51funmanS_a_i_n_t: you know who created it / updated it the last time ?
21:11:00*kugel wonders if the glyph cache is needed in RaaA
21:11:20S_a_i_n_tfunman: No sorry, all I know is that it hasn't been updated in an age.
21:12:26 Quit t0rc (Remote host closed the connection)
21:12:43S_a_i_n_tat least bilditup1 has reminded me to put the CygWin Wiki page on my ToDo list.
21:12:46 Quit bilditup1 (Read error: Connection reset by peer)
21:12:50S_a_i_n_tIt is pretty outdated.
21:13:09S_a_i_n_t(re-reminded me I should say)
21:13:29 Join bilditup1 [0] (
21:14:45 Quit jfc (Ping timeout: 260 seconds)
21:15:45funmanat least, the arm compiler isn't up to date on this image
21:16:47 Quit Jerom (Quit: Leaving.)
21:17:29S_a_i_n_tfunman: From what I recall...someone recently kludged a new image together.
21:17:34S_a_i_n_tCheck the forums.
21:17:40 Join Jerom [0] (~heidi@
21:17:42S_a_i_n_tI seem to remember this being the case.
21:17:52funmanwhich subforum/thread?
21:18:20S_a_i_n_tPass...sorry, I usually just check "recent posts".
21:18:39B4gderI don't think it matters who made the previous one
21:18:42B4gderanyone can make a new
21:18:55B4gderit just takes someone to actually do it
21:19:40funmanB4gder: it does matter
21:20:05n1sthere's this
21:20:11funmanmotivating this person to make a new image is less work than making a new image
21:20:18funmans/is/could be/
21:20:24B4gdera new image is a new image to anyone who makes it
21:20:47S_a_i_n_tmay, or may not, help.
21:20:48n1sisn't it just updating the toolchain and posting the new image?
21:20:56B4gderI would guess so
21:21:11funmanB4gder: depends if this person just has to run apt-get update /
21:21:18funmanand in any case it is less work for me
21:21:23B4gderfunman: but the image is the same image you download
21:21:28B4gderwhich is the same to everyone
21:21:48funmanah so i could just run it, run some commands and distribute it again, right?
21:22:03funmanok so motivating this person is less work for me
21:22:17lestatarhi all, got a quick quest re: pictureflow and gigabeatF40 please :-)
21:22:21funmando not disregard my laziness :)
21:22:33bieberbluebrother: What problems are you having with it?
21:22:37B4gderhehe, well I'm not doing it either ;-)
21:22:49lestatarlong press in WPS screen on fuze v2 shows a pictureflow option, but on gigabeat pictureflow does not appear...
21:22:51lestatarany ideas?
21:22:51mc2739funman: thanks, how does midi_debug_printf sound?
21:23:14S_a_i_n_tlestatar: Check Hotkey, in the manual.
21:23:20S_a_i_n_tThis is a hotkey feature
21:23:34lestatargreat thanks saint :-)
21:23:45funmanmc2739: longer than "midi_debug" ;)
21:23:46 Join jfc [0] (
21:25:18 Join Kitr88 [0] (
21:25:41CIA-8New commit by 03bluebrother (r27561): Fix USB ID retrieval on Windows 7. ...
21:26:48bluebrotherbieber: I've fixed it a bit earlier, but something is different to the first import. The main problem is that the qtfindreplacedialog code assumes to get compiled as a library, and thus adds __declspec(dllimport) / __declspec(dllexport) prefixes to functions on Windows.
21:27:13CIA-8r27561 build result: All green
21:27:30 Quit arcnaut (Quit: Ex-Chat)
21:27:34bieberIt was originally packaged as a library, but I stripped all the library export stuff out of it and just added the classes to my project directly
21:27:48bieberI must have missed some of it, then
21:27:49bluebrotherthe old version also contained less files. Checking upstream didn't show much activity in the last month so this was a bit surprising.
21:27:58bluebrotherah, that explains things :)
21:28:23bluebrotherwell, it's working now and I've also created new windows binaries. Just in case someone wants to try ;-)
21:28:25kugelany reason to not keep it as a library?
21:28:46bluebrotheryes: we want a static binary.
21:28:59 Quit Kitar|st (Ping timeout: 276 seconds)
21:29:06kugelthe theme editor too?
21:29:21bluebrother(of course we could adjust it so the library builds as static lib but that's ... well, work :)
21:29:31bieberkugel: For something that's just a couple of forms, I'd rather not turn it into a whole separate library
21:29:53bluebrothersure, it makes deployment much easier. Unless we provide an installer.
21:30:14 Quit Kitr88 (Ping timeout: 264 seconds)
21:30:19bluebrotherin which case packaging is more complex.
21:30:52kugeli see
21:31:32kugelis qt also statically linked?
21:31:34bluebrotherI've looked in creating an installer for Windows recently, and have something thats kinda working. It _might_ be worth considering to use non-static builds for that.
21:31:42bluebrotherif you build it statically, yes :)
21:32:48kugelI'm not sure if I did that, but I guess it works either way (because it did when I compiled it on my desktop)
21:32:55kugelnice work btw :)
21:33:29bluebrotherbieber: I've managed to get NSIS download the fonts pack and extract it, currently into a subfolder fonts/ in the installation directory. Have you thought about having the theme editor fallback to a specified location for fonts?
21:34:18bieberIt will always fall back to the fonts directory that's set in the preferences dialog
21:34:43 Join Kitar|st [0] (
21:34:47bluebrotherif you build on Linux you have Qt installed and no need to build statically. The problem begins when deploying a binary: we don't know if the user has Qt on his machine at all. So using a static build is a good thing here.
21:35:21CIA-8New commit by 03mc2739 (r27562): Rename printf to prevent naming conflict. Also change comment to conform with Rockbox standards.
21:35:31kugelisn't the usual fix to provide the the .dll nevertheless?
21:35:35bluebrotherOn Windows you could simply place the Qt dlls in the same location as the application, but that means that you don't have a single binary anymore and need to extract the complete archive to some location.
21:35:57bluebrotherwhat's the fix about that?
21:36:03bieberOn Linux platforms, though, we can package it for the various distros so that the package manager will install Qt
21:36:26lestatarhi joy with the hotkey suggestion
21:36:30kugelwell, fix/workaround; this is the reason everbody has a couple dozen copies of the same dll
21:36:37bluebrotheryou have to ship the dll in some way. Either built in (i.e. statically linked) or as separate file. This works fine on Windows
21:36:48lestatari might not have explained my quest correctly...
21:37:02lestatarpictureflow does not appear in the context menu of the WPS screen
21:37:05bluebrotherhowever, on Linux the directory the user extracted the binary to is extremely likely to not get searched for libs.
21:37:06CIA-8r27562 build result: 280 errors, 160 warnings (mc2739 committed)
21:37:25bluebrotherwhich means that one needs a (simple but yet existing) shell script around the binary to adjust that.
21:37:42bluebrothera statically compiled binary is easier here. At least from the deployment point of view.
21:38:40bluebrotheron Linux the package manager can handle the dependencies, so no problem here. Unless we provide a binary that is simply a binary and not a distro specific package. Then we get the same issue as with Rockbox Utility.
21:39:26bieberEven if it's just a binary, the standard on unix systems is generally to use shared libraries, and put the dependencies in the README, no?
21:39:32bluebrotherbieber: NSIS is one of the common installers on Windows. See
21:39:45bluebrotherno, I've never seen that.
21:41:28n1s<n1s> if i open a file and overwrite it's entire content with something else but with the same size, is the old content guaranteed to be overwritten?
21:41:43n1s<n1s> i'm thinking about implementing some kind of schredding for keybox, when importing info from a text file
21:42:41bluebrotherbieber: well, my idea with the fonts dir was that you can have the "standard" fonts already installed without setting a fallback path. Or the other way round: consider the "standard" fonts part of the installation (once / if we get an installer), and the user might want to add more fonts. The application folder might not be writable by the user so a separation is needed here.
21:43:35kugeln1s: IIRC yes if you add O_TRUNC to the flags
21:43:43bieberShouldn't non-standard fonts go in the project's fonts/ directory, though? Since they're going to need to be exported with the project anyways
21:43:53S_a_i_n_tlestatar: Perhaps it is yet to be implemted for the gigabeat.
21:43:58n1skugel: thanks
21:44:08S_a_i_n_tOr, perhaps you need to update the build on your gigabeat.
21:44:10 Join bunnyboi [0] (
21:44:24bluebrotherhmm. Yes, but wouldn't it make sense to have a "common fonts" folder (where the font is taken from when exporting the theme)?
21:44:36 Quit bunnyboi (Read error: Connection reset by peer)
21:44:53bluebrotherOTOH, that use case might be neglectibly small.
21:44:57lestatarah, works running from demo plugins was installed via rb util 3.6 the other day so pretty recent
21:45:32lestatarone other question please...
21:46:00lestatari have around 3500 songs on the gigabeat, but over 7k db this correct?
21:46:53n1slestatar: that sounds wrong to me, each file should have one entry
21:47:09CIA-8New commit by 03mc2739 (r27563): Replace remaining printf in midi plugin (fix yellow/red)
21:47:17bieberbluebrother: That would be my thought. The two separate directories would be doable, of course, but I feel like it would introduce a lot of confusion with distribution. If we stick the "common" fonts in the program directory, then you'd have to run it with root permissions to update them on unix
21:47:23 Quit BHSPitMini (Ping timeout: 265 seconds)
21:47:29lestataryeah, i believe this is remnanat of encrypted sat files which were on the device when i first installed rb...
21:47:46lestatari have since deleted all those files from the drive
21:47:52lestatarhow best should i rebuild my db?
21:48:24bluebrotherbieber: I agree. However, the common path could be differently set on Windows and Linux (assuming that we install the fonts via an installer on Windows ;-)
21:48:25n1slestatar: just "Initialize Now" in the menu
21:48:45CIA-8r27563 build result: All green
21:48:48n1s(will nuke any "runtime data" though)
21:49:02lestatarthanks n1s...i did that previously, but still have the extra entries...
21:49:22lestatarwill try again tho and dont care about the runtime data
21:49:22bluebrotherthe assumption to have write permissions to the program installation path might be wrong on Windows too. Hmm.
21:49:39 Quit funman (Quit: free(random());)
21:50:42n1slestatar: the most common cause for extra entries are file still there in a OS trashcan thingy
21:50:46 Quit bilditup1 (Ping timeout: 265 seconds)
21:50:55bluebrotherlestatar: as far as I can see from the logs Pictureflow WPS integration was committed after 3.6 was released.
21:51:08bluebrotherso in that sense 3.6 is not "pretty decent" at all :)
21:51:12lestatarthansk guys... picture flow...thanks!
21:51:45lestatarthe recycle bin was emptied on gigabeat in explorer
21:52:00lestatartho there is still a restore folder in the bin...
21:52:04 Quit merbanan (Ping timeout: 258 seconds)
21:52:16lestatarshould i delete that as well? is this possibly cause of my extra db entries?
21:52:42 Join AndyI [0] (~pasha_int@
21:54:21n1slestatar: does it have music files in it?
21:54:25bluebrotherthat restore folder _could_ still contain the files.
21:54:48lestatarone sec, lemme connect it...
21:56:22lestatarok, this gigabeat is used and was a bit of a mess
21:56:36lestatarmsc shows recycled bin empty
21:56:59lestatarthe restore folder is under system volume info folder and is also empty
21:57:29lestatarhowever, under gbsystem/image, there are 2088 files with .gbp extension
21:57:38lestatarno idea what these are...
21:57:54lestatarshould i delete? i don't care about them as i only have music on the device
21:58:29 Quit antil33t (Read error: Connection reset by peer)
22:02:40bluebrotherbieber: btw, if you want to map nicknames to real names check the IrcNicks wiki page :)
22:03:29lestatarah, never mind guys... .gbp files are created by gigabeat, 3 for every image under the picture folder...dates are very old [prior to me] to i am gonna delete them
22:03:47lestatarhope this will allow initialize db to remove the extra entries
22:04:48 Join merbanan [0] (
22:05:36***Saving seen data "./dancer.seen"
22:05:51 Join antil33t [0] (
22:12:25lestatarok, getting closer! :-) just about 4000 songs and now 5275 db entries...
22:13:16lestataru also have a temp data folder with some avi file i am storing on the device...this is close enuff...everything works perfectly anyway
22:13:18lestatarthank you all! :-D
22:18:49 Quit milz (Read error: Connection reset by peer)
22:19:02 Join milz [0] (
22:19:25*n1s spots something that looks very fishy
22:24:16bertrikn1s, what is it?
22:27:03n1satrac3_rm.c: 38
22:28:26bertrikpfff, looks nasty indeed
22:30:38bertrikmaybe the same trick was used when putting the data in id3v2buf first?
22:32:58bertrikin apps/metadata/oma.c:154 the buffer is filled with AV_WL16(&id3->id3v2buf[0], 1); // always 1
22:34:10bertrikwait, I'm confusing some structs I think
22:35:21n1sit is indeed, metadata/rm.c:411
22:35:42n1sbut is this struct guaranteed not to move?
22:36:08n1shmm, wait, if it moves, this should still work
22:38:16n1ssince the codec memcpy's the struct to a local var, alignment shouldn't be a big problem
22:38:48kugelgevaerts: there's some trickery needed to make themes from work with the app, do you think it's worth it?
22:40:12 Quit B4gder (Quit: It is time to say moo)
22:40:23bertrikn1s, indeed
22:40:53*kugel decides to worry about that later
22:41:29 Quit Jaykay (Remote host closed the connection)
22:43:49n1sbertrik: hmm, removing that pointer magic hade no difference though
22:43:49bertrikn1s, but it's at least not getting worse?
22:43:49n1sso at least it probably isn't causing my bug :)
22:43:49 Join mt_ [0] (~mtee@
22:44:43bertrikIf code points to structs in id3v2buf (which is not a good idea IMO), I think it would be better to make sure that id3v2buf is aligned instead of having alignment hacks in the codecs
22:45:30lestatarhey folks...just an fyi...current build r27563 for gigabeat f/x causes themes to not work for some reason...
22:45:46lestatarbut indeed pictureflow is part of context menu :-)
22:46:00lestatari am rolling back to the previous build
22:46:07kugelwhy are structs in id3v2buf?
22:50:26bertrikkugel, I'm not really familiar with this part of rockbox but I think this buffer is a kind of general purpose storage thing between the metadata parser and the codec
22:51:48gevaertskugel: the .rockbox name?
22:52:15mt_kugel: Some metadata parsers use id3v2buf to store the extracted data instead of parsing the file again in the codec itself.
22:52:34 Nick mt_ is now known as mt (~mtee@
22:52:51gevaertskugel: maybe rbutil can be made to unpack them right
22:54:03kugelthat could work too
22:54:25kugelit's not exactly huge trickery to make rockbox convert "/.rockbox" though
22:54:56kugelI think I got it to work a few minutes ago, but for some reason I'm having problems with the themes (that use ui vp) so I cannot really verify
22:56:31mtCould someone with a CF target test something ? (needs test_codec and wmapro)
22:57:54n1smt: sure, btw, did you see FS #11495 ?
22:58:17mtn1s: Nope was out of town, just came back.
22:58:37n1sbertrik: i agree, aligning the buffer is nicer
22:58:59mtn1s: It's a very small patch that adds the output buffer to iram
22:59:02 Join mischasworld [0] (
22:59:43n1smt. ok
23:00:19 Join robin0800 [0] (~robin0800@
23:01:35 Quit S_a_i_n_t (Read error: No route to host)
23:09:37 Quit milz (Ping timeout: 240 seconds)
23:10:54 Join S_a_i_n_t [0] (S_a_i_n_t@
23:11:48 Quit mt (Ping timeout: 258 seconds)
23:12:16n1smt: 63.89% realtime -> 64.17% realtime for the 55kbps sample, not worth it imo
23:12:45 Join mt_ [0] (~mtee@
23:12:53 Nick mt_ is now known as mt (~mtee@
23:14:21mtn1s: Thanks
23:16:39CIA-8New commit by 03bieber (r27564): Theme Editor: Added recent docs/projects menus. Modified buildtargetdb.php to add a do-not-modify warning to the top of its output, and generated a ...
23:16:47n1smt: that buffer isn't very hot, it seems the processing is done more on the buffers in the globWMAProDecCtx struct
23:16:49 Quit mischasworld (Remote host closed the connection)
23:17:38n1sthe tmp buffer for example
23:18:09CIA-8r27564 build result: All green
23:18:18n1si'll test that
23:19:26mtn1s: yeah you're right, that was a poor choice.
23:20:26merbananer are you working on a wmapro decoder for rockbox =
23:20:50 Join bunnyboi [0] (
23:21:03mtmerbanan: Yes
23:21:34merbananneat :)
23:23:48merbananmt: are you doing the work ?
23:24:23gevaertsHe'd better be!
23:25:20mtmerbanan: Yes It currently just needs optimisations to work realtime on coldfire and still no multichannel support.
23:28:51n1smt: putting the tmp buffer in iram is slightly better 64.61% realtime but not worth it either imo, we need to find what buffer is hottest
23:29:07mtprobably coeffs
23:29:13mtn1s ^
23:29:38 Quit dockimble (Ping timeout: 260 seconds)
23:29:45n1sone worthwhile optimization would be to get rid of the sample interleaving since the rockbox dsp has very efficient code for that anyway
23:29:50mtn1s: Sorry I mean "out"
23:30:14n1syeah, out seems hot too
23:30:23n1si'll try that
23:32:52 Join Buschel [0] (
23:34:15n1sthe interleaving also copies all samples an extra time since pcmbuf_insert will copy them anyway
23:34:53mtI'm working on the interleaving part now
23:36:47n1sgreat :)
23:40:14 Join panni__ [0] (
23:41:05 Quit robin0800 (Quit: Leaving)
23:41:57n1sout buffers don't fit in iram even for stereo
23:42:22n1sat least on mcf5249 (48k of codec iram)
23:43:09 Quit panni_ (Ping timeout: 260 seconds)
23:45:33 Quit ender` (Quit: /[,\(\s]?#1(s)(\s+-?\d+(?{}\.\d+)?\s+-?\d+(?{}\.\d+)?\s+-?\d+(?{}\.\d+)?\s+-?\d+(?{}\.\d+)?\s+-?\d+(?{}\.\d+)?\s+-?\d+(?{}\.\d+)?)(\s+-?\d+(?{}\.\d+)?\s+-?\d+(?{}\.\d+)?)+(\)?=|\s|$)/)
23:46:58n1sthose out buffers are 16k each and the mdct tables+code already take up a portion of that iram
23:49:21Buschelbut it should work for codecs with more iram (e.g. mcf5250 or pp502x) −− if we lower the maximum number of channels to stereo...
23:49:26 Join stripwax [0] (
23:50:17 Quit Strife89 (Ping timeout: 240 seconds)
23:51:21Buschelor we just use iram'ed buffers for the first two channel (stereo) and use non-iram'ed buffers for the next channels. this would help for the standard use case.
23:52:03 Quit VoivoD (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716])
23:52:03 Join sudoman [0] (
23:53:40n1swe can also have one channel's buffer in iram and the other in dram for stereo
23:54:07n1shmm, no they're in fact 24k each, not 16
23:54:44Buschelyep, just wanted to state that. they are huge.
23:55:38Buschelbut not a problem for pp502x and mcf5350 ;)
23:55:48Buschelmcf5250, of course
23:55:59 Quit bunnyboi (Quit: +++ OK ATH OK)
23:56:17 Join DerPapst [0] (
23:57:48n1syeah, with one of those buffers in iram the 55kbps sample is 70.6% realtime so a relative speedup of ~10%

Previous day | Next day